On the passivity-based power control of a doubly-fed induction machine

I. Lpez-Garc a
a
, G. Espinosa-Prez
a,,1
, H. Siguerdidjane
b
, A. Dria-Cerezo
c,2
a
Universidad Nacional Autnoma de Mxico, FIUNAM, Apartado Postal 70-256, 04510 Mxico DF, Mexico
b
Automatic Control Department, Suplec, Plateau du Moulon, 3 rue Joliot-Curie, 91192 Gif-sur-Yvette, Cedex, France
c
Universitat Politcnica de Catalunya, DEE and IOC, EPSEVG, Av. V. Balaguer s/n, Vilanova i la Geltr 08800, Spain
a r t i c l e i n f o
Article history:
Received 5 September 2011
Received in revised form 24 August 2012
Accepted 29 August 2012
Available online 23 October 2012
Keywords:
Double-fed induction machines
Passivity-based control
Power regulation
a b s t r a c t
In this paper the stator-side power regulation control problem of DFIM is approached. It is shown how a
passivity-based controller can deal with this problem establishing a viable solution from both a dynamic
performance perspective and a practical implementation point of view. The evaluation of the presented
scheme is carried out by considering typical operation conditions, namely: active power generation with
demanded or delivered reactive power. Special interest is given to the rigorous establishment of the sta-
bility properties of the closed-loop system, which allows for achieving remarkable dynamic responses, in
terms of convergence rates as well as the region of attraction of the equilibrium point dened by a given
operation regime. From a practical viewpoint, the usefulness of the scheme is concluded by realizing that
the requirements imposed concerning the structural features of both the generator by itself and the
power converter required for its operation, can be fullled by commercially available off-the-shelf
devices.
2012 Elsevier Ltd. All rights reserved.
1. Introduction
Doubly-fed induction machines (DFIMs) have become a funda-
mental element in several applications related with efcient meth-
ods for electric energy generation. This fact is due, on the one hand,
to the high performance that can be achieved with this kind of
devices under variable-speed operation and, on the other hand,
to the possibility for feeding directly the rotor windings using high
efcient power converters, which makes feasible to carry out the
energy conversion process using only a small fraction of the energy
managed by the whole power system [13].
Motivated by these remarkable features, DFIM has been deeply
studied and several useful properties are currently well-known, for
example, that if the model is represented in a line-voltage vector-
oriented reference frame, then the controller design can be carried
out in a simpler and more robust way, since the stator active and
reactive power can be controlled in a decoupled way while the
coordinate transformation for representing the model in the afore-
mentioned reference frame is parameter-independent [4].
Starting with the pioneering work of [1], currently several con-
trollers have been proposed in the literature, specially in applica-
tions related with wind generation, e.g. [5,6]. Unfortunately, in
most of these contributions both the design and/or the perfor-
mance analysis rely on the application of classical linear control
techniques [8,9], simplifying model assumptions (either consider-
ing linearized models [7], negligible some of the machine parame-
ters [10] or using reduced order models [11]) or by implementing
non-robust open-loop integration methods for estimation of
unmeasurable variables (e.g. stator uxes [12]), while (to the best
of the authors knowledge) only a few references can be found,
where a rigorous complete stability analysis is presented, being
the results presented in [13,14] some illustrative examples.
The purpose of this paper is to contribute to the proposition of
control schemes for DFIM that can deal with the stator-side power
regulation control problem with rigorously established stability
properties. The motivation for contributing in this way to the solu-
tion of this problem lies in the conviction that better structured
control schemes leads to the achievement of higher performances.
To this aim, in this paper it is shown that the controller presented
by the authors in [15], where it was considered the case of having
an isolated load, can be directly implemented to solve the power
control problem, in the sense that some constant reference for
the stator-side active and reactive powers can be reached asymp-
totically. This objective is achieved by identifying, for some given
power reference, the corresponding values for stator and rotor cur-
rents and mechanical speed that must be achieved and showing
that the considered control scheme indeed states a stabilization
0142-0615/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijepes.2012.08.067

Corresponding author.
E-mail addresses: irvinlopez@yahoo.com (I. Lpez-Garc a), gerardoe@unam.mx
(G. Espinosa-Prez), Houria.Siguerdidjane@supelec.fr (H. Siguerdidjane), arnau.
doria@upc.edu (A. Dria-Cerezo).
1
On sabbatical leave at LSS-SUPELEC, France, supported by SUPELEC, DGAPA-
UNAM (IN111211), II-FI-UNAM (Grant 1111) and Universit Paris-Sud XI.
2
Arnau Dria-Cerezo was partially supported by the Spanish government research
project DPI2010-15110.
Electrical Power and Energy Systems 45 (2013) 303312
Contents lists available at SciVerse ScienceDirect
Electrical Power and Energy Systems
j our nal homepage: www. el sevi er . com/ l ocat e/ i j epes
mechanism for this operation conditions, provided the mechanical
torque delivered to the generator is constant.
The usefulness of the controller scheme is evaluated assuming
the more typical operation conditions for the DFIM, i.e. active
power is always fed into the grid (for different values of stator
power factor) while it is allowed both to deliver or demand reac-
tive power to/from the grid. This different scenarios are considered
by letting the constant mechanical torque to take different values.
The achieved stability and performance properties of the closed-
loop system are illustrated, rst, under an stringent scenario,
where the generator is at standstill at the beginning of the exper-
iments, and, second, under a more practical situation, where the
operation departs from non-zero initial conditions. The attractive-
ness of the control scheme is concluded from two different per-
spectives, namely: From the dynamic point of view, due to the
stability properties and performance exhibited by the closed-loop
system, and from a practical viewpoint, since the magnitude of
all the system variables correspond to commercially available
off-the-shelf experimental setups (including both the generator
and the power converter).
The rest of the paper is organized as follows: In Section 2, the
considered model for the DFIM is presented, together with the con-
trol problem formulation and its solvability analysis. The proposed
controller is developed in Section 3 while its evaluation is carried
out in Section 4. Section 5 is devoted to state some concluding
remarks.
2. Problem formulation
In this section the considered model for the DFIM is presented
to later on state the approached control problem and the condi-
tions under which it is solvable.
2.1. DFIM model
Under the assumption of linear magnetic circuits and balanced
operating conditions, the equivalent two-phase model of the sym-
metrical DFIM, represented in a rotating dq reference frame xed
to the stator voltage vector, is given by [4]
di
s
dt
x
s
Ji
s
xbJk
r
ci
s
abk
r

bL
r
L
sr
u
s
bu
r
1
_
k
r
x
s
xJk
r
aL
sr
i
s
ak
r
u
r
2
J
_
x
L
sr
L
r
i
T
s
Jk
r
BxT
m
3
where x
s
is the rotation speed for the reference frame, x is the
rotor speed, i
s
= [i
s1
, i
s2
]
T
are the stator currents, k
r
= [k
r1
, k
r2
]
T
are
the rotor uxes, u
s
and u
r
are the stator and rotor voltages,
respectively, while the all positive parameters are given by
a
R
r
L
r
; b
L
sr
l
; c
1
l
R
s
L
2
r
R
r
L
2
sr
L
r
_ _
with l L
s
L
r
L
2
sr
and
J
0 1
1 0
_ _
J
T
Here L
s
, L
r
are stator and rotor proper inductances, L
sr
is the mutual
inductance, R
s
, R
r
are the winding resistances, J is the inertia coef-
cient, B is the damping coefcient and T
m
is the applied mechanical
torque.
Considering the ux vector k L
e
i with k k
T
s
; k
T
r
_
T
and
i i
T
s
; i
T
r
_ _
T
, where k
s
are the stator uxes and i
r
the rotor currents,
while
L
e

L
s
I
2
L
sr
I
2
L
sr
I
2
L
r
I
2
_ _
; I
2

1 0
0 1
_ _
4
model (1)(3) can be equivalently written as
_
k
s
x
s
L
s
Ji
s
x
s
L
sr
Ji
r
R
s
i
s
u
s
5
_
k
r
x
s
xL
sr
Ji
s
x
s
xL
r
Ji
r
R
r
i
r
u
r
6
J
_
x L
sr
i
T
s
Ji
r
BxT
m
7
Remark 1. Under generator operation of the DFIM, T
m
is the torque
delivered to the machine by a controlled primary mover while
T
g
L
sr
i
T
s
Ji
r
is torque produced by the machine itself. In this paper
it is considered that T
m
is constant, assumption that is not
restrictive since usually the primary mover is equipped with a
speed controller [16].
Remark 2. From a controller design perspective, one advantage
exhibited by the DFIM is that the complete state vector is measur-
able, i.e. mechanical speed and both stator and rotor currents can
be used to structure of the control scheme.
2.2. Power control problem
From the generated power viewpoint, model (5)(7) exhibits
the following structure if it is assumed that the stator terminals
are connected to an innite bus with voltage magnitude U and fre-
quency determined by x
s
. Active P and reactive Q power at the
stator side are given by
P
ab
I
T
s
V
s
; Q
ab
I
T
s
JV
s
where I
s
and V
s
are the vectors of stator currents and voltages,
respectively, in the natural ab reference frame for the induction
machine.
3
In the dq reference frame considered for representing
the machine model, these expressions take the form
P Ui
s1
; Q Ui
s2
8
which clearly exhibit the advantage of the representation, since
control of the active and reactive power can be carried out by con-
trolling each of the components of the stator current vector i
s
.
Taking into account the information presented above, the con-
trol problem solved in this paper can be stated as.
Consider the DFIM model given by (5)(7). Assume that
A.1 The mechanical speed and both stator and rotor currents are
available for measurement.
A.2 The torque delivered by the prime mover is constant and
known.
A.3 The model parameters are known.
A.4 Stator voltages have xed frequency and amplitude, i.e.
stator windings are directly connected to the line grid.
Under these conditions, design a control law for the rotor
voltages u
r
= u
r
(i
s
, i
r
, x) such that
lim
t!1
P P
I
; lim
t!1
Q Q
I
with the desired power dened by P
I
, Q
I
, guaranteeing internal
stability.
3
In the ab reference frame the stator variables are represented in a xed reference
frame while the rotor variables are expressed with respect to a reference frame that
rotates at an angular speed given by x [4].
304 I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312
2.3. Solvability analysis
Once the control problem has been stated, this section is
devoted to analyze under which conditions it is possible to nd a
viable solution. This analysis is carried out by studying the behav-
ior of the machine operating in steady state conditions with the
aim to identify the required values that must be reached by the
state variables of the system in order to achieve the desired oper-
ation. From a stability viewpoint, this procedure corresponds to
identify, among the different equilibrium points of the dq model,
the assignable equilibrium points, i.e. those that satisfy the im-
posed conditions. From a controller design perspective, identifying
these points has the additional advantage of establishing the
desired operation point that must stabilized in order to succeed
in the solution of the control problem.
The equilibria of system (5)(7) are determined by the set of
algebraic equations given by
x
s
Ji
sI
x
s
L
sr
Ji
rI
R
s
i
sI
u
s
0 9
x
s
x
I
L
sr
Ji
sI
x
s
x
I
L
r
Ji
rI
R
r
i
rI
u
rI
0 10
L
sr
i
T
sI
Ji
rI
Bx
I
T
m
0 11
where the subindex q stands for the value of the variables under
equilibrium operation. Notice that under the stated assumptions
u
s
and T
m
remain constant.
The rst point to be noticed is that, due to the existence of the
control input given by the rotor voltages u
r
, Eq. (10) can always be
satised by a proper choice of this variable. Thus, the analysis must
be concentrated in Eqs. (9) and (11). On the other hand, from Eq.
(8) it is clear that for a prescribed value of the active and reactive
powers, given by P
I
and Q
I
respectively, the corresponding stator
currents are given by
i
sI

1
U
1 0
0 1
_ _
P
I
Q
I
_ _
12
since u
s
= [U, 0]
T
.
With the desired values for the stator currents, it is easy to
obtain, from (9), the value of the rotor current i
rq
that must be
achieved as
i
rI

1
x
s
L
sr
J
T
x
s
J R
s
I
2
i
sI
u
s

1
x
s
L
sr
x
s
I
2
R
s
Ji
sI
J
T
u
s
13
where it was used the fact that J
T
J = JJ
T
= I
2
.
The nal step, which is related with the computation of the
required rotor speed x
q
, is carried out by substituting Eq. (13) into
(11) to obtain that
x
I

1
Bx
s
i
T
sI
Jx
s
I
2
R
s
J
T
i
sI
J
T
u
s

T
m
B

1
Bx
s
P
I
R
s
ki
sI
k
2

T
m
B
14
since JJ = I
2
and ki
sI
k
2
i
T
sI
i
sI
, with k k the Euclidean norm.
Even that from a mathematical perspective Eqs. (12)(14)
clearly state the conditions that must be satised in order to
achieve a given power in the stator side of the machine, from a
practical point of view several conditions must be taken into
account in order to solve the stated control problem, namely:
C.1 Given a desired power behavior (in terms of a given i
sq
) it
must be considered the capacity of the machine concerning
the nominal rotor current that can be managed, i.e. careful
selection must be done in order to avoid demanding a rotor
current beyond the limits imposed by the construction of the
DFIM. This constraint is also related with the size of the
power converter required to operate the machine.
C.2 Eq. (14) exhibits the power balance that must be satised by
the machine. In this sense, for a given torque T
m
, selection of
the desired active power (and the associated losses) will
determine the speed under which the machine must oper-
ate. In the same spirit than the mentioned in the previous
condition, the limits imposed by the physical structure of
the DFIM must be taken into consideration.
3. Stabilization mechanism
In this section it is presented the proposed controller that ren-
ders asymptotically stable the assignable equilibrium point intro-
duced in Section 2.3. The design is carried out under a passivity-
based approach [17,18], i.e. by exploiting the energy dissipation
properties of the machine, due to the attractive results previously
obtained solving power systems related problems, e.g. [1921].
The section starts by briey recalling the rational behind the design
approach to continue with the presentation of the controller.
3.1. Passivity-based control
The problem of stabilizing an equilibrium point of nonlinear
systems of the form
_ x f x; t gxu
where x 2 R
n
is the state vector, u 2 R
m
(m < n) is the control action
and g(x) is assumed full rank, is approached from the Interconnec-
tion and Damping Assignment-Passivity Based Control (IDA-PBC)
perspective by nding a control law u(x) that leads to a closed-loop
system of the form
_ x F
d
x
@H
d
x
@x
15
where F
d
x F
T
d
x 6 0 and H
d
(x) P0 is a scalar (energy) function,
which satises the condition
x
I
arg min H
d
x
with x
q
the equilibrium to be stabilized.
4
A system with structure like the introduced in (15) is known as
a Hamiltonian system, and the aim of achieving such structure is
that it enjoys several properties [23]. Among them, one that is
quite important from the stability point of view is related with
the fact that if H
d
(x) is considered as Lyapunov function of the sys-
tem, then its time derivative along the trajectories of (15) is given
by
_
H
d

@H
d
x
@x
_ _
T
F
d
x F
T
d
x
_ _
@H
d
x
@x
allowing for the possibility of proving the asymptotic stability prop-
erties of x
q
[22], due to the negativeness of F
d
x F
T
d
x.
A particular case, that will be exploited in this paper, is when
H
d
(x) is linear, i.e. is a quadratic function of the form
H
d
x
1
2
x
T
Px
with P = P
T
> 0 a symmetric positive denite matrix. Under these
conditions
@H
d
x
@x
Px
4
Notice that the gradient of the scalar function H
d
(x), given by
@Hdx
@x
, is considered
as a column vector throughout the paper.
I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312 305
and therefore
_
H
d
x
T
P F
d
x F
T
d
x
_ _
Px
guaranteeing automatically the asymptotic stability of x
q
if
F
d
x F
T
d
x < 0.
3.2. Controller design
In order to apply the IDA-PBC approach, it is convenient at this
point to recall that the total energy stored by the DFIM is given by
Hz
1
2
z
T
L
1
z
where L diagfL
e
; Jg while z = [k
T
, Jx]
T
. Thus, a natural candidate
for the desired energy function of the closed-loop system is
H
d
z
1
2
z z
I

T
P
d
z z
I

with P
d
P
T
d
> 0 and z
I
k
T
I
; Jx
I
_
T
, where k
I
L
e
i
I
. As required,
this function has a minimum in z z
q
= 0, i.e. when k = k
q
and
x = x
q
, the former implying that the minimum is achieved when
i = i
q
.
If the matrix P
d
has the following structure
P
d

p
s
I
2
0 0
0 p
r
I
2
0
0 0 p
m
_

_
_

_
with p
s
, p
r
and p
m
positive constants (to be dened latter) and the
matrix F
d
(z) is partitioned in a conformal way as
F
d
z
F
11
z F
12
z F
13
z
F
21
z F
22
z F
23
z
F
31
z F
32
z F
33
z
_

_
_

_
then the desired closed-loop system (15) takes the form
_ z F
d
zP
d
z z
I
F
d
zP
d
Le 16
with
e
e
s
e
r
e
m
_

_
_

_
i
s
i
sI
i
r
i
rI
xx
I
_

_
_

_
which (after carrying out all the required computations) can be
written as
_
z
L
s
p
s
F
11
L
sr
p
r
F
12
L
sr
p
s
F
11
L
r
p
r
F
12
Jp
m
F
13
L
s
p
s
F
21
L
sr
p
r
F
22
L
sr
p
s
F
21
L
r
p
r
F
22
Jp
m
F
23
L
s
p
s
F
31
L
sr
p
r
F
32
L
s
p
s
F
31
L
sr
p
r
F
32
Jp
m
F
33
_

_
_

_e 17
In order to nd the structure of F
d
(z) and the values for p
s
, p
r
and
p
m
that satisfy the required conditions, consider the rst constraint
given in terms of the stator ux dynamics (5) as
x
s
Ji
s
x
s
L
sr
Ji
r
R
s
i
s
u
s
L
s
p
s
F
11
L
sr
p
r
F
12
e
s
L
sr
p
s
F
11
L
r
p
r
F
12
e
r
Jp
m
F
13
e
m
This equation, under the value obtained for u
s
from the equilib-
rium Eq. (9) can be written as
x
s
J R
s
I
2
e
s
x
s
L
sr
Je
r
L
s
p
s
F
11
L
sr
p
r
F
12
e
s
L
sr
p
s
F
11
L
r
p
r
F
12
e
r
Jp
m
F
13
e
m
leading directly to the solution given by F
33
= 0 and
F
11
F
12
_ _

L
s
p
s
I
2
L
sr
p
r
I
2
L
sr
p
r
I
2
L
r
p
r
I
2
_ _
1
x
s
J R
s
I
2

x
s
L
sr
J
_ _
18
Concerning the second constraint, obtained from the rotor ux
dynamics (6), the presence of the control input u
r
allows for the
possibility of satisfying the corresponding equation no matter the
value of F
d
(z), p
s
, p
r
and p
m
. In fact, taking advantage of this facility,
an useful choice for the corresponding components of the matrix
F
d
(z) (whose usefulness will be better recognized during the stabil-
ity analysis) is given by
F
21
F
T
12
; F
23
F
T
32
; F
22

k
r
2p
r
I
2
19
with k
r
a gain (to be dened later).
Once the second constraint has been studied, the attention
must be concentrated in the third one, imposed by (7) and given as
L
sr
i
T
s
Ji
r
BxT
m
L
s
p
s
F
31
L
sr
p
r
F
32
e
s
L
s
p
s
F
31
L
sr
p
r
F
32
e
r
Jp
m
F
33
e
m
Again, exploiting the value obtained for T
m
from the equilibrium
Eq. (11), this last expression can be presented as
L
sr
i
T
s
Ji
r
L
sr
i
T
sI
Ji
rI
Be
m
L
s
p
s
F
31
L
sr
p
r
F
32
e
s
L
s
p
s
F
31
L
sr
p
r
F
32
e
r
Jp
m
F
33
e
m
from which it can immediately recognized the necessity for having
F
33

B
Jp
m
and leading to the expression
L
sr
i
T
s
Ji
r
L
sr
i
T
sI
Ji
rI
L
s
p
s
F
31
L
sr
p
r
F
32
e
s
L
s
p
s
F
31
L
sr
p
r
F
32
e
r
After substitution of the errors e
s
and e
r
in the left-hand side of
this last expression and manipulating it in a proper way, it can be
rewritten as
e
T
s
e
T
r
_
T
p
s
F
T
31
p
r
F
T
32
_ _
L
sr
L
1
e
Ji
rI
Ji
s
_ _
_ _
0
from, where a possible solution for F
31
and F
32
could be obtained by
letting the term in the brackets equal to zero. Unfortunately, this
solution is not useful since F
31
would depend on the state i
s
, feature
that (as will be clear later) does not allow for achieving the desired
stability properties. Therefore it is necessary to write this condition
in an equivalent way given by
e
T
s
e
T
r
_
T
p
s
F
T
31
p
r
F
T
32
_ _
L
sr
L
1
e
Ji
rI
Ji
s
_ _
Gz
_ _
0
provided that
e
T
s
e
T
r
_
T
Gz e
T
s
e
T
r
_
T
G
1
z
G
2
z
_ _
0
Under these conditions, if
Gz

Ls L
2
sr
l
Je
s

L
3
sr
l
Je
r

L
3
sr
l
Je
s

Lr L
2
sr
l
Je
r
_
_
_
_
with l as in the DFIM model, then
F
31

L
sr
lp
s
k
T
rI
J; F
32

L
sr
l
k
T
s
J 20
are solutions of the constraint.
Until this point, putting together the results obtained in (18)
(20), it has been obtained a matrix F
d
(z) that satises Eq. (16) for
the given structure of P
d
. Indeed, this matrix is given by
F
d
z

1
p
s
x
s
J
Lr Rs
l
I
2
_ _
Lsr Rs
lp
r
I
2
0

Lsr Rs
lp
r
I
2

kr
2p
r
I
2
Lsr
l
J
T
k
s
Lsr
lp
s
k
T
rI
J
Lsr
l
k
T
s
J
B
Jp
m
_

_
_

_
21
306 I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312
With matrix (21) at hand, then the remaining task is to nd the
conditions for guaranteeing that F
d
z F
T
d
z 6 0. To this end, con-
sider that under the structure of F
d
(z) given above, it is obtained
that
F
d
z F
T
d
z

2Lr Rs
lp
s
I
2
0
Lsr
lp
s
J
T
k
rI
0
kr
p
r
I
2
0
Lsr
lp
s
k
T
rI
J 0
2B
Jp
m
_

_
_

_
where the advantage of choosing the skew-symmetric terms in the
design of F
d
(z), in particular the feature of leaving the element F
33
depending only on the constant value k
rq
, is now evident, since
using the standard tool given by the Schur complement it is possi-
ble to prove that the required condition F
d
z F
T
d
z < 0 is achieved
if
p
s
>
JL
2
sr
4BlL
r
kk
rI
k
2
_ _
p
m
22
Under the condition found for the parameters p
s
and p
m
, all the
elements that are necessary for implementing the control law are
at disposition. Then from (6) and the corresponding row of (16),
it is possible to state the following
Result 1. Consider the DFIM model (5)(7) and assume that A.1 to
A.4 hold. Let the rotor control input be given as
u
r
x
s
xJL
sr
i
s
L
r
i
r
R
r
i
r
p
s
F
21
L
s
e
s
L
sr
e
r

p
r
F
22
L
sr
e
s
L
r
e
r
Jp
m
F
23
e
m
23
where
F
21

L
sr
R
s
lp
r
I
2
; F
22

k
r
2p
r
I
2
; F
23

L
sr
l
J
T
L
s
i
s
L
sr
i
r

while k
r
> 0, p
r
> 0, p
m
> 0 and
p
s
>
JL
2
sr
4BlL
r
kL
sr
i
sI
L
r
i
rI
k
2
_ _
p
m
Dene the equilibrium point of the closed-loop as
i
sI

1
U
1 0
0 1
_ _
P
I
Q
I
_ _
i
rI

1
x
s
L
sr
x
s
I
2
R
s
J
T
i
sI
J
T
u
s

x
I

1
Bx
s
P
I
R
s
ki
sI
k
2

T
m
B
Under these conditions, it is achieved that
lim
t!1
P P
I
; lim
t!1
Q Q
I
guaranteeing internal stability.
Remark 3. As pointed out in [15], the structure of the presented
controller can be further simplied, from a tuning point of view,
by dening
k
s

p
s
L
sr
R
s
lp
r
; k
m

p
m
JL
sr
lp
r
since, under these conditions, the controller takes the form
u
r
x
s
xJL
sr
i
s
L
r
i
r
R
r
i
r
k
s
L
s
e
s
L
sr
e
r
k
r
L
sr
e
s
L
r
e
r

k
m
JL
s
i
s
L
sr
i
r
e
m
with k
r
> 0, k
m
> 0 and
k
s
>
L
2
sr
4BlL
r
kL
sr
i
sI
L
r
i
rI
k
2
_ _
k
m
4. Controller evaluation
The usefulness of the presented control scheme is analyzed in
this section. This analysis is divided into two parts, namely: First,
Table 1
DFIM parameters.
R
s
= 4.92[X]
R
r
= 4.42[X]
L
s
= 0.725[H]
L
r
= 0.715[H]
L
sr
= 0.71[H]
J = 0.00512[kg m
2
]
B = 0.005[N m/rad/s]
0 0.1 0.2 0.3
1
0
1
2
x 10
4
P
s

[
W
]
Time [s]
0 0.1 0.2 0.3
4
3
2
1
0
1
x 10
4
Q
s

[
V
A
R
]
Time [s]
0 0.1 0.2 0.3
5
0
5
10
15
20
x 10
4
P
r

[
W
]
Time [s]
0 0.1 0.2 0.3
2
0
2
4
6
8
x 10
4
Q
r

[
V
A
R
]
Time [s]
Fig. 1. Stator and rotor powers for PI 1750:7W, Q
I
0VAR with zero initial
conditions.
0 1 2 3 4 5 6
0
50
100
150
200
250
w

[
r
a
d
/
s
]
Time [s]
Fig. 2. Mechanical Velocity for PI 1750:7W, QI 0VAR with zero initial
conditions.
I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312 307
the stability properties of the control law are numerically evalu-
ated. It is illustrated how, as predicted by the theory, all the vari-
ables converge to their desired values both, under an extreme
operation regime (zero initial conditions) and under a more realis-
tic condition (non-zero initial conditions). Second, for several oper-
ation regimes the corresponding values for the rotor and speed
variables are given with the aim to evaluate if the requirements
imposed by the controller operation are feasible to achieve in a
practical setting. This evaluation procedure was carried out under
the MATLAB/Simulink environment.
Throughout all the operation conditions, there were considered
the same machine parameters, which are included in Table 1, the
same controller gains k
s
= 1700, k
r
= 1000 and k
m
= 0.18, while the
innite bus was modeled in such a way that the amplitude of the
stator terminal voltages was xed at U = 220[V].
To illustrate the stability (convergence) features exhibited by
the control law, it is rst considered the case when it is demanded
some amount of active power in the stator side with zero reactive
power. With the aim to evaluate the scheme under drastic condi-
tions, it was assumed that the generator was at standstill (all the
initial conditions were set to zero). In this case, the reference for
the stator active power was P
I
1750:7W with a desired stator
power factor PF
sq
= 1 while the mechanical torque was kept at
T
m
= 5[N m]. From Figs. 14 the stator and rotor powers, the
mechanical speed, the stator and rotor currents and a phase
voltage of the stator and rotor are shown, respectively. In all these
pictures it can be shown how the desired behavior is achieved in a
very short time. It is also clear that a transient response of consid-
erable magnitude is exhibited, but this behavior is due to the
stringent initial conditions imposed to the generator, as will be
clear in the next experiments when a more practical situation is
considered.
A second experiment, as well as the rst one more oriented to
stress the stability properties of the control scheme, is presented
from Figs. 58. In this case the reference for the stator active power
was P
I
1400:6W while for the reactive power was
Q
I
1050:4VAR, with a desired reactive power of PF
sq
= 0.8
0 0.1 0.2 0.3
20
10
0
10
20
30
i
s
d

[
A
]
Time [s]
0 0.1 0.2 0.3
20
0
20
40
60
i
s
q

[
A
]
Time [s]
0 0.1 0.2 0.3
30
20
10
0
10
20
i
r
d

[
A
]
Time [s]
0 0.1 0.2 0.3
60
40
20
0
20
i
r
q

[
A
]
Time [s]
Fig. 3. Stator and rotor currents for PI 1750:7W, Q
I
0VAR with zero initial
conditions.
0 2 4 6
215
220
225
230
V
s
d

[
V
]
Time [s]
0 2 4 6
2
1
0
1
2
V
s
q

[
V
]
Time [s]
0 2 4 6
0
100
200
300
400
500
V
r
d

[
V
]
Time [s]
0 2 4 6
2000
1500
1000
500
0
500
V
r
q

[
V
]
Time [s]
Fig. 4. Stator and rotor voltages for PI 1750:7W, Q
I
0VAR with zero initial
conditions.
0 0.1 0.2 0.3
1
0
1
2
x 10
4
P
s

[
W
]
Time [s]
0 0.1 0.2 0.3
4
3
2
1
0
1
x 10
4
Q
s

[
V
A
R
]
Time [s]
0 0.1 0.2 0.3
5
0
5
10
15
20
x 10
4
P
r

[
W
]
Time [s]
0 0.1 0.2 0.3
0
2
4
6
8
x 10
4
Q
r

[
V
A
R
]
Fig. 5. Stator and rotor powers for PI 1400:6W, QI 1050:4VAR with zero
initial conditions.
0 1 2 3 4 5 6
0
50
100
150
200
250
300
350
400
w

[
r
a
d
/
s
]
Time [s]
Fig. 6. Mechanical Velocity for PI 1400:6W, Q
I
1050:4VAR with zero
initial conditions.
308 I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312
and considering a mechanical torque of T
m
= 5[N m]. In the same
way than in the rst experiment, the convergence properties of the
controller are well illustrated. Again, the same remark than in the
previous experiment concerning the considerable transient
response applies.
With the aim to further evaluate the controller scheme consid-
ering not only its stability but also its performance properties, the
two additional experiments were carried out now considering a
more realistic operation. To this end, instead of considering the
limit case of having zero initial conditions, it was considered that
the system operates under some given conditions and a step
change in the required power is imposed, i.e. different from zero
initial conditions were considered. In the rst case, it was required
that the stator active power passes from an initial value of
P
I
2100W to a nal one of P
I
1750W while the reactive
power changes from an initial value of Q
I
300VAR to a desired
value of Q
I
0VAR. The desired behavior considered a unitary
power factor and the mechanical torque was set to T
m
= 5[N m].
Fig. 9 presents the stator and rotor powers evolution, where it can
be noticed how the required values are achieved in around 3 s,
while in Fig. 10 the corresponding mechanical speed is exhibited.
In Fig. 11 both the stator and rotor currents are included. In these
last three pictures, it can be noticed that although the currents
exhibit some transient response, its magnitude is now reasonable
in a practical context, as well as the required speed. This remark
is also applicable to the demanded rotor voltages which are
included in Fig. 12 together with the stator voltages.
The second case under this non-zero initial conditions scenario,
was devoted to the case when the nal operation conditions
involve a non-zero reactive power value. In this case the initial
conditions were set in such a way that generator was operating
at P
I
1250W with Q
I
1150VAR while the nal desired
value for the power was P
I
1400W with Q
I
1050VAR.
The considered power factor was PF
sq
= 0.8 and the mechanical
torque was T
m
= 5[N m]. The dynamic behavior of the generator
variables are exhibited in Fig. 13 for the stator and rotor powers,
Fig. 14 for the mechanical speed, Fig. 15 for the stator and rotor
currents and Fig. 16 for the stator and rotor voltages. In all of them
it can be noticed how the desired behavior is achieved under
reasonable transient values for all the variables.
0 0.1 0.2 0.3
20
10
0
10
20
30
i
s
d

[
A
]
Time [s]
0 0.1 0.2 0.3
20
0
20
40
60
i
s
q

[
A
]
Time [s]
0 0.1 0.2 0.3
30
20
10
0
10
20
i
r
d

[
A
]
Time [s]
0 0.1 0.2 0.3
60
40
20
0
20
i
r
q

[
A
]
Time [s]
Fig. 7. Stator and rotor currents for PI 1400:6W, QI 1050:4VAR with zero
initial conditions.
0 2 4 6
215
220
225
230
V
s
d

[
V
]
Time [s]
0 2 4 6
2
1
0
1
2
V
s
q

[
V
]
Time [s]
0 2 4 6
200
0
200
400
V
r
d

[
V
]
Time [s]
0 2 4 6
2000
1500
1000
500
0
500
V
r
q

[
V
]
Time [s]
Fig. 8. Stator and rotor voltages for PI 1400:6W, Q
I
1050:4VAR with zero
initial conditions.
0 2 4 6
2500
2000
1500
1000
P
s

[
W
]
Time [s]
0 2 4 6
1000
500
0
500
1000
Q
s

[
V
A
R
]
Time [s]
0 2 4 6
500
1000
1500
2000
2500
P
r

[
W
]
Time [s]
0 2 4 6
0
500
1000
1500
Q
r

[
V
A
R
]
Time [s]
Fig. 9. Stator and rotor powers for PI 1750W, QI 0VAR with non-zero
initial conditions.
0 1 2 3 4 5 6
40
60
80
100
120
140
160
180
200
220
w

[
r
a
d
/
s
]
Time [s]
Fig. 10. Mechanical Velocity for PI 1750W, Q
I
0VAR with non-zero initial
conditions.
I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312 309
The second part in the evaluation of the presented controller
includes several operation conditions that are presented in a
compact way in Table 2. Concerning the procedure followed for
carrying out this part of the evaluation, it was assumed that the
generator always fed active power into the grid while there were
considered two conditions for the reactive power: when the DFIM
operates as a capacitor (delivering reactive power to the grid) and
when it operates as an inductive load (demanding reactive power
from the grid). In all these cases the system was simulated for dif-
ferent values of the power factor including both lagging and lead-
ing values. To achieve these operation conditions, as in the
previous experiments, once the desired value for power (P
I
and
Q
I
) was dened, the corresponding components of the desired sta-
tor currents were computed to later on obtain the corresponding
desired rotor current and speed, i
rq
and x
q
respectively. Looking
for an uniform evaluation context, the components of the desired
stator current vector were dened in such a way that the magni-
tude of this vector was kept at the same value in all the experi-
ments, i.e. the condition that ki
sq
k
2
= 2.6[A] always was satised.
In addition, variation of the delivered mechanical torque was also
considered.
In the rst six columns of Table 2 are presented the required
behavior in terms of the applied mechanical torque, the desired
power factor, () Lagging, (+) Leading, the desired power (active
and reactive) and the xed conditions considered for the innite
bus. The second part is devoted to present the conditions that
are necessary to match in the rotor side of the machine in order
to achieve the desired conditions. Here negative values of power
represent active or reactive power supplied by the stator or the
rotor and positive values represent the amount of power that they
demand. The voltages and currents correspond to the rms magni-
tude of one of the phases while the last column stands for the slip
frequency, i.e. the difference between the actual and the synchro-
nous speeds in steady-state. It must be noticed that after conver-
gence of the actual variables to the desired ones, the values
included for the stator and rotor currents and the mechanical
speed, correspond to the computed from Eqs. (18)(20), neverthe-
less, they are included for completeness purposes.
0 2 4 6
3.5
3
2.5
2
1.5
i
s
d

[
A
]
Time [s]
0 2 4 6
1
0.5
0
0.5
1
i
s
q

[
A
]
Time [s]
0 2 4 6
1.5
2
2.5
3
3.5
i
r
d

[
A
]
Time [s]
0 2 4 6
2
1.5
1
0.5
0
i
r
q

[
A
]
Time [s]
Fig. 11. Stator and rotor currents for PI 1750W, Q
I
0VAR with non-zero
initial conditions.
0 2 4 6
215
220
225
230
V
s
d

[
V
]
Time [s]
0 2 4 6
2
1
0
1
2
V
s
q

[
V
]
Time [s]
0 2 4 6
50
100
150
200
250
V
r
d

[
V
]
Time [s]
0 2 4 6
10
0
10
20
30
40
V
r
q

[
V
]
Time [s]
Fig. 12. Stator and rotor voltages for PI 1750W, Q
I
0VAR with non-zero
initial conditions.
0 2 4 6
1700
1600
1500
1400
1300
1200
P
s

[
W
]
Time [s]
0 2 4 6
1300
1200
1100
1000
900
800
Q
s

[
V
A
R
]
Time [s]
0 2 4 6
500
400
300
200
100
0
P
r

[
W
]
Time [s]
0 2 4 6
800
600
400
200
Q
r

[
V
A
R
]
Time [s]
Fig. 13. Stator and rotor powers for PI 1400W, QI 1050VAR with non-
zero initial conditions.
0 1 2 3 4 5 6
360
370
380
390
400
410
420
430
w

[
r
a
d
/
s
]
Time [s]
Fig. 14. Mechanical Velocity for PI 1400W, Q
I
1050VAR with non-zero
initial conditions.
310 I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312
The usefulness of the information included in Table 2 is mainly
concentrated in the information concerning the rotor side of the
machine. It can be viewed from two different perspectives. The
rst one is related with conditions C.1 and C.2 introduced in Sec-
tion 2.3. In this sense, all the values obtained for the motor vari-
ables could be used to identify safe operation regimes in terms of
the nominal characteristics of the rotor side when an specic DFIM
is analyzed. The information can be used to conclude if it is possi-
ble or not to achieve a given amount of desired generated power.
The second point of view refers to the possibility of using this
information from a design perspective. If the analysis is carried
out without considering any specic machine, then the informa-
tion could be used to carry out a proper selection of both, the gen-
erator by itself and the size of the converter that must be
implemented for feeding the rotor windings.
In the particular numeric example considered in Table 2, from
the information provided by the column that displays the mechan-
ical speed, it can be concluded that for a proper operation of the
scheme it is not necessary to consider something else but a stan-
dard prime mover, since the speed values required under the sev-
eral operation conditions can be reached with a standard device for
the corresponding power requirements.
Concerning the power converter that must be considered for
equipping the generator, its structure is mainly dened by the
power that it must handle and the frequency at which it must
operate, i.e. the information contained in the last three columns
of Table 2. In this context it is important to recognize that having
positive and negative values of the slip frequency (which means
the necessity for operating at both sub and super-synchronous
speeds) implies that the converter must be able to change the rotor
phase sequence. Moreover, since there are also positive and nega-
tive (active and reactive) rotor powers, then it will be also required
a bidirectional operation for this device. Fortunately, these charac-
teristics are easily fullled with standard back to back converters
and the only point that is necessary to study is the size that is
related with the switching frequency operation and the power
capacity. However, even from this perspective, the required values
for all the operation regimes can be easily managed with a
standard converter of reasonable size.
The only point that must be treated with more detail is related
with the required rotor current. As can be seen in the correspond-
ing column of the table, the required value of this variable in-
creases when both active and reactive power is delivered to the
grid. This fact generates the necessity for considering a DFIM capa-
ble of working under this demanded current. However, even that
this is the major limitation of the scheme, achieving values of
0 2 4 6
2.6
2.4
2.2
2
1.8
i
s
d

[
A
]
Time [s]
0 2 4 6
1.4
1.6
1.8
2
i
s
q

[
A
]
Time [s]
0 2 4 6
1.8
2
2.2
2.4
2.6
i
r
d

[
A
]
Time [s]
0 2 4 6
3.2
3
2.8
2.6
2.4
2.2
i
r
q

[
A
]
Time [s]
Fig. 15. Stator and rotor currents for PI 1400W, Q
I
1050VAR with non-
zero initial conditions.
0 2 4 6
215
220
225
230
V
s
d

[
V
]
Time [s]
0 2 4 6
2
1
0
1
2
V
s
q

[
V
]
Time [s]
0 2 4 6
80
60
40
20
V
r
d

[
V
]
Time [s]
0 2 4 6
25
20
15
10
V
r
q

[
V
]
Time [s]
Fig. 16. Stator and rotor voltages for PI 1400W, QI 1050VAR with non-
zero initial conditions.
Table 2
DFIM response to xed stator power exchange.
T
m
[N-m] PF
sq
P
sq
[W] Q
sq
[VAR] i
sa
[A] u
sa
[V] i
ra
[A] u
ra
[V] x [rad/s] P
r
[W] Q
r
[VAR] f
r
[H]
4.5 0.8 1400.6 1050.4 2.6 220 2.28 47.48 261.2 322.9 43.2 8.42
5 0.8 1400.6 1050.4 2.6 220 2.28 23.43 361.2 156.1 38.4 7.49
5.5 0.8 1400.6 1050.4 2.6 220 2.28 94.17 461 635.2 120.1 23.41
4.5 0.9 1575.6 763.1 2.6 220 2.47 102.63 186.9 761.6 22.2 20.25
5 0.9 1575.6 763.1 2.6 220 2.47 30.55 286.9 226.8 4.7 4.33
5.5 0.9 1575.6 763.1 2.6 220 2.47 41.52 386.9 308 12.7 11.48
4.5 1.0 1750.7 0 2.6 220 2.90 162.28 112.6 1302 557.4 32.07
5 1.0 1750.7 0 2.6 220 2.90 87.66 212.6 711.8 280.8 16.16
5.5 1.0 1750.7 0 2.6 220 2.90 13.89 312.6 121.2 4.2 0.24
4.5 0.9 1575.6 763.1 2.6 220 3.28 108.08 186.9 823.3 675.3 20.25
5 0.9 1575.6 763.1 2.6 220 3.28 32.76 286.9 288.5 144.5 4.33
5.5 0.9 1575.6 763.1 2.6 220 3.28 46.49 386.9 246.2 386.3 11.58
4.5 0.8 1400.6 1050.4 2.6 220 3.41 51.30 261.2 407.9 330.7 8.42
5 0.8 1400.6 1050.4 2.6 220 3.41 29.55 361.2 71.1 294.3 7.49
5.5 0.8 1400.6 1050.4 2.6 220 3.41 104.62 461.2 550.2 919.2 23.41
I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312 311
around 3.5[A
rms
] can be considered yet under a reasonable limit for
a DFIM as the considered for the numerical evaluation.
5. Concluding remarks
It was shown in this paper that the controller proposed in [15]
can be directly applied in order to solve the stator-side power reg-
ulation control problem of DFIM. This conclusion is valid for both, a
dynamic performance scenario and a practical implementation
perspective. The procedure followed for evaluating the control
scheme involved typical operation conditions for a generator of
the approached kind and the obtained results allowed for the illus-
trating the stability (convergence) properties of the control law,
including stringent (zero initial conditions) and more practical
conditions (non-zero initial conditions). In addition, it was shown
that for all the operation regimes the structural requirements for
the generator as well as for the power converter required to oper-
ate the machine, correspond to standard devices that can be easily
obtained in a practical context. It remains to develop a deeper
research work with the aim to relax the assumption of having a
constant mechanical torque delivered by the prime mover in order
to enlarge the possible applications of the proposed controller.
References
[1] Leonhard W. Control of electric drives. Springer; 1995.
[2] Kumar NS, Gokulakrishnan J. Controller on the stability of power systems
connected with doubly fed induction generator. Int J Electr Power Energy Syst
2012;33:117284.
[3] Ghedamsi K, Aouzellag D. Performances for wind energy conversion systems.
Int J Electr Power Energy Syst 2010;32(9):9356.
[4] Krause CP, Wasynczuk O, Sudhoff DS. Analysis of electric machinery and drive
system. 2nd ed. IEEE Press Power and Engineering Series; 2002.
[5] Boukhezzar B, Siguerdidjane H. Nonlinear control with wind estimation of a
DFIG variable speed wind turbine for power capture optimization. Energy
Convers Manage 2009;50:88592.
[6] Chen CH, Hong CM, Cheng FS. Intelligent speed sensorless maximunm power
point tracking control for wind generation system. Int J Electr Power Energy
Syst 2012;42:399407.
[7] Rahimi M, Parniani M. Dynamic behavior analysis of doubly-fed induction
generator wind turbines: the inuence of rotor and speed controller
parameters. Int J Electr Power Energy Syst 2010;32(5):46477.
[8] Wang J, Tse N, Gao Z. Synthesis on PI-based Pitch controller of large wind
turbine generator. Energy Convers Manage 2011;52(2):128894.
[9] Yang L, Xu Z, Ostergaar J, Yang Z, Wong KP, Ma X. Oscillatory stability and
eigenvalue sensitivity analysis of a DFIG wind turbine system. IEEE Trans
Energy Convers 2011;26(1):32839.
[10] Pal BC, Mei F. Modelling adequacy of doubly fed induction generator for
small-signal stability studies in power systems. IET Renew Power Generat
2008;2(3):18190.
[11] Mei F, Pal BC. Modelling of doubly fed induction generator for power system
stability study. IEEE Power Energy Gen Meet 2008:18190.
[12] Xu L, Cartwright P. Direct active and reactive power control of DFIG for wind
energy generation. IEEE Trans Energy Convers 2006;21(3):7508.
[13] Peresada S, Tilli A, Tonielli A. Power control of a doubly fed induction machine
via output feedback. Control Eng Practice 2004;12(1):4157.
[14] Monroy A, Alvarez-Icaza L, Espinosa-Prez G. Passivity based control for
variable speed constant frequency operation of a DFIG wind turbine. Int J
Control 2008;81(9):1399407.
[15] Batlle C, Dria-Cerezo A, Espinosa-Prez G, Ortega R. Simultaneous
interconnection and damping assignment passivity-based control: the
induction machine case study. Int J Control 2009;82:24155.
[16] Munteanu I, Bratcu AI, Cutululis NA, Ceanga E. Optimal control of wind energy
systems: towards a global approach. Springer; 2008.
[17] Ortega R, van der Schaft AJ, Mareels I, Maschke B. Putting energy back in
control. IEEE Control Syst Mag 2001;21(2):1833.
[18] Ortega R, Garcia-Canseco E. Interconnection and damping assignment
passivity-based control: a survey. Eur J Control 2004;10:43250.
[19] de Leon J, Espinosa-Perez G, Macias I. Observer-based control of a synchronous
generator: a hamiltonian approach. Int J Electr Power Energy Syst
2002;24(8):65563.
[20] Manjarekar NS, Banavar RN, Ortega R. Application of interconnection and
damping assignment to the stabilization of a synchronous generator with a
controllable series capacitor. Int J Electr Power Energy Syst 2010;32(1):6370.
[21] He B, Zhang X, Zhao X. Transient stabilization of structure preserving power
systems with excitation control via energy-shaping. Int J Electr Power Energy
Syst 2007;29(10):82230.
[22] Khalil H. Nonlinear systems. Prentice Hall; 2002.
[23] van der Schaft AJ. L
2
-Gain and passivity techniques in nonlinear
control. Springer-verlag; 2000.
312 I. Lpez-Garc a et al. / Electrical Power and Energy Systems 45 (2013) 303312